The demands made on the technology are enormous: it simply has to be better and more reliable than humans. How can this work?
To illustrate, highly-automated driving is about nothing less than replacing human beings and all their knowledge, experience, judgment, and their ability to act and react behind the wheel. What's more, technology has to improve. Errors that human drivers are likely to make are unacceptable in highly-automated driving.
Which Sensors Are Necessary?
Environmental sensors play an important part in highly-automated driving. They are the equivalent of a human’s senses installed into a self-driving car. The focus is on three systems: cameras, radar and lidar.
Cameras are the closest thing there is to the human eye. 2D cameras can detect infrastructure when driving, i.e. lines, signs, and objects. They also distinguish between trucks, cars, pedestrians and cyclists in detail. Thanks to their ability to recognize colors, cameras are essential for responding to traffic signals.
Four fisheye cameras with a wide aperture angle of 180 degrees – installed on all four sides of the vehicle – also enable optimum detection of the vehicle’s complete environment at close range. These cameras are supplemented by a front camera integrated into the base of the rear-view mirror, with a smaller aperture angle and a wider range for detecting pedestrians and vehicles.
Simply put, a radar emits electromagnetic rays that are reflected by humans and objects. The distance and speed of other traffic participants can thus be determined by measuring the time of the reflection. Radars can be categorized according to their range: long-range radars such as the Advanced Radar Sensor are used, for example, in the adaptive cruise control. The radar detects the vehicle ahead, measures distance and speed, and brakes the vehicle's own speed accordingly. Short-range radars have a smaller range but larger aperture angle to scan the immediate surroundings of the vehicle and are used, for example, during parking and for monitoring the “blind spot”.
High Resolution 3D Flash LIDAR
LIDAR systems emit light pulses in the infrared range and measure the time required for the pulse to return. With 3D Flash LIDAR (LINK TO ARTICLE “3D Flash LIDAR”), which measures in several places at the same time, higher resolutions are possible than with radar systems. In addition, they recognize the outlines of objects and the surface texture of roads.
How Does the Technology Deal With Complex Situations?
Despite all the ongoing development of individual systems. Simple systems alone are not enough to replace people flawlessly. Imagine a highway construction site with narrow lanes, road lines in different colors worn down and ripped, vehicles packed close together, and oncoming traffic moving past you in the very next lane. It’s a real challenge but usually not a problem for a skilled driver. You take your lead from the vehicle in front, ignore a ripped road line and despite all the confusion, you figure out how to get through the roadworks.
And the technology? The solution lies in redundancy: only when the various sensors interact can a reliable and complete detection of the environment be possible. More than a dozen environmental sensors are used. This sensor information is supplemented with highly precise map material and more details from the back end. Finally, all information is collected in a central control unit and processed. By drawing on all this data and information, the control unit can accurately depict the vehicle’s surroundings in an environment model.